Enterprise Analytic Applications: A Guide to the Latest Developments
By Timo ElliottSenior Director of Strategic Marketing, Business Objects
Ever since mainframe computers began accumulating vast storehouses of data in the early 1960s, managers and executives have sought ways to turn random facts and figures into useful information upon which to base sound business decisions.
But it wasn't until the advent of relational databases and client/server technology in the early 1990s that companies took advantage of the market's need for decision support systems to create and define a new industry, which is now widely known as business intelligence (BI). Business intelligence allows organizations to extract useful, actionable information from a rapidly growing inventory of disparate data sources, including multiple database platforms, packaged applications, data warehouses, data marts and e- business systems.
The major database and enterprise-class packaged application vendors typically supply basic querying and reporting functionality for their own products. Some argue this has led to a "stovepipe" model of information in the organization - isolated, myopic islands within a corporate structure that accumulate and warehouse their own data but share it reluctantly, if at all, with the rest of the company. BI vendors provide tools that can be used across the organization to access, analyze and share information from a variety of sources - a giant step in the right direction for business decision-makers needing the big picture.
As the use of BI has matured, there has been increased interest in analytic applications, a logical extension of the business intelligence concept. Analytic applications provide users with prepackaged solutions to common business problems such as customer, sales and campaign analysis. Although analytic applications have been available in areas such as financial budgeting for many years, they typically relied on older, proprietary systems and covered only a small fraction of the overall needs of the enterprise. Over the last few years, analytic applications have gained popularity in new areas, particularly for the analysis of e- business and clickstream information.
Analytic applications provide key additional BI benefits to specific groups of end users through the use of "best practice" analysis techniques - in particular business areas and "closed loop" integration with operational systems. However, their implementation has not always been painless. While they address a business need for a particular population, they perpetuate the problem of stovepipe information sources and may make it more difficult than ever to get an overall view of the enterprise. As Gartner noted in a May 2001 report: "Packaged BI applications may seem appealing in the context of a particular application, but organizations should ensure that they will support BI in a broader context as a strategic initiative."
Hence the notion of "enterprise analytic applications" which provide a common platform for analytic applications throughout the organization. The situation is similar to the changes that have taken place in the packaged operational applications market, where companies first implemented packaged solutions rather than creating them from scratch (PeopleSoft for human resources, for example) then quickly demanded integrated systems that tied together operations across the organization, pioneered by companies such as SAP and rapidly followed by the other vendors.
The chief business benefits of an enterprise analytic applications approach are:
With these constants in mind, let's examine the evolutionary path of enterprise analytic applications, including where they are today, where they're going and how to create a BI strategy that works.
The Origins of Analytic Applications
Analytic applications offer prebuilt solutions to complex business problems. They are typically shared applications that support multiple users in which the interaction is programmed by the application vendor and where, in addition to standard business intelligence functionality (access, analyze, share) they provide systems to process data into information required to answer strategic business questions. For example: "Who are our most profitable customers?"
Given this definition, the first early analytic applications came on the market in the late 1970s with the advent of the first multidimensional databases. These systems were first used for finance and budgeting and later - when they became more robust - for sales and marketing.
These systems continued to gain popularity over the next two decades, but were never widely deployed outside a small group of core analysis users, and the underlying technology generally failed to keep up with client/server and Internet technologies.
But for many reasons, analytic applications are now emerging as a key trend in business intelligence. The mainstream implementation of enterprise resource planning applications such as software from SAP, PeopleSoft and Oracle have streamlined business workflows within organizations, but, some argue, they have by and large provided little beyond basic reporting. Frustrated end users want to get at more of the valuable information stored within these systems.
The second reason is improving technology. While relational databases provided powerful storage for a wealth of information, they have typically not been able to rival the capacity of multidimensional databases for complex analysis. New technologies, including set-based analysis, are reducing these barriers by allowing robust analytic applications to be built on the more advanced and more standard relational platform.
Finally, the maturation of the business intelligence market has also driven the trend towards analytic applications. As organizations deploy BI tools more widely - and vendors gain more experience of "typical" deployments in an increasing variety of different industry and functional areas - the benefits to packaging and making these solutions readily available are becoming obvious.
Benefits of Analytic Applications
At a high level, the primary benefit of an analytic application is the ability to simplify analysis that would otherwise require a complex series of steps. Making analysis easier lowers one of the key barriers to the widespread use of BI. Rather than business users relying on intuition (or guessing) when making decisions, analytic applications can help make analysis an automatic part of the business process.
In The Loyalty Effect by Frederick Reichheld, the author demonstrates that even a small increase in customer retention rates (from 90 percent to 95 percent) can result in a big increase in profits (more than 50 percent in this example). Using business intelligence, a marketing manager might want to identify the most loyal customer segments - and what percentage of profits came from new customers, frequent customers buying more and existing customers buying less.
But this type of analysis is surprisingly difficult for a business user in most corporate environments, and therefore is only rarely carried out. The way data is typically stored in relational databases and data warehouses requires examining each customer's purchases line by line to define the segment containing "customers that bought more." This is a process that may result in thousands of separate database queries. Instead of wrestling with this complexity, most organizations continue to work using much- higher level segments. In turn, they may miss key underlying trends in customer data.
Take the real-life example of a dot- com company at the height of the Internet boom. The site's marketing managers were content because their analysis showed that their key revenue segment of customers had grown by more than 100 percent in the past year. As a result, their strategies focused primarily on finding ways to further improve the acquisition of these valuable customers.
But a more detailed analysis by loyalty segment showed that instead of growing at 100 percent, this key segment had in fact increased at more than 600 percent over the year. But the analysis also revealed that most of the new customers had also left the segment. In other words, the company was already doing an extremely good job of acquiring new customers, but a very poor job of retaining them - a problem they didn't spot because of inadequate customer analytics. With this analysis in hand, they were able to better target their marketing efforts to a retention strategy.
Analytic applications embed technology that makes creating and pre-calculating useful customer segments such as these much easier. For example, complex analysis can be done with a single database query instead of thousands of queries. These applications also place analysis under the direct control of the end user, without the need for intervention from the information technology department. Users can, for example, go on to look at, create and analyze other segments such as "customers who bought less" and determine what potential profits might have been without customer turnover.
Another key benefit of these applications is best-practice analytics. For example, a customer intelligence analytic application might contain pre-defined routines based on customer segmentation and segment migration analysis techniques that represent the cutting edge of customer relationship management. Thus, analytic applications help end users with the most fundamental and vital steps in the BI process, prompting them on which questions to ask and directing them toward the proper techniques for a specific type of analysis.
Finally, because analytic applications are designed for a specific business or functional area, the results of the analysis can be tied directly back to the operational systems for "closed loop analysis." For example, the results of an analysis of soon-to- churn customers can be used to drive an e-mail marketing campaign using a packaged CRM application.
Overall, analytic applications hold the promise of bringing organizations closer to "repeatable business excellence." Just as packaged applications have helped streamline and standardize company operations, analytic applications help organizations make consistent decisions based on all the relevant data. Businesses can move ahead faster without having to relearn what others already know. This saves them time and money - and helps them strengthen their competitive position.
Pitfalls When Implementing Analytic Applications
We've seen that analytic applications represent a valuable, logical next step in the deployment of business intelligence. But they also come with several pitfalls.
The most insidious pitfall is that unless analytical applications are carefully implemented, organizations can end up with a jumble of different "stovepipe" BI applications using different or loosely integrated technologies and data structures. Rather than contributing to a company's overall business intelligence IQ, these stovepipes make the implementation of a more comprehensive BI solution even more difficult to create and sustain.
How do these stovepipes come about? In some cases, they are a natural extension of a precise need - to do clickstream analysis of the organization's web site, for example. In other cases, some argue, analytic applications are provided by the large packaged application vendors who make it financially tempting to purchase the reporting and analytics tools that correspond with their different applications.
The problem here, some contend, is that most organizations have a combination of different packaged applications. These packaged applications rarely cover all the various operations of an organization. According to Gartner: "Through 2005, broad-scale adoption of packaged applications will prevent more than 50 percent of large organizations from establishing complete perspective through BI." As a countermeasure, Gartner recommends implementing an application- neutral data warehouse.
Enterprise Analytic Application Implementation Strategies
This leads directly to the need for "enterprise analytic applications," which are a series of analytic applications that respond to the user's needs for best-practice analytics, while maintaining a common framework that fits with the rest of the company's data warehousing and BI strategy.
Ideally, these applications are data-source neutral, able to take information from anywhere within the organization, and integrate with existing query and reporting tools within the company.
Since there will always be operations that are specific to a particular company or industry, the ideal enterprise analytic applications model allows for a "build-and-buy" approach. The main analytic needs are covered by predefined applications - customer intelligence, supply chain intelligence - while niche applications could be created internally, using the same platform and infrastructure. This means that the data stored by the organizations can still be effectively combined and compared as part of an overall BI strategy.
This approach brings organizations closer to the ideal of "one truth" - a consistent enterprise view across different elements of the business, different business units, or different departments within a business unit. Users need one truth. For example, total revenue in the sales system exactly equals total revenue in the product line revenue reporting system - a holistic view of corporate performance measurement.
In most industries, key measures such as product, channel and customer profitability cannot be correctly calculated by domain-specific applications. Instead, data from multiple domains, such as finance and CRM, need to be processed, typically using activity-based costing techniques, before profitability can be accurately calculated. Once these measures are in place, organizations can drive the business and compensate employees on appropriate, consistent metrics - for example, measuring marketing departments not just on the number of leads that have been brought in, but on the total profit brought in by those leads over a given period of time.
There are also clear cost gains to be achieved by avoiding unnecessary BI product proliferation. By setting standards and deploying a single architecture, IT organizations can save money in selection, implementation, training and maintenance.
Analytic applications provide clear benefits for business users by simplifying complex analysis and providing best practice measures and techniques. But the benefits of implementing these applications can be outweighed by the drawbacks of "stovepipe" data stores unless organizations implement the applications as part of an overall BI roadmap.
Enterprise analytic applications provide the benefits of analytic applications without stovepipes by using a data-neutral approach. This provides a common platform for both querying and reporting, as well as a suite of predefined best practice BI applications. These applications move organizations closer to the "holy grail" of one truth, where each department within the organization is measured by the same clearly defined, relevant standards of business performance, while eliminating the costs associated with supporting a jumble of non-integrated BI solutions.
To achieve this, key business users must work closely with the IT personnel responsible for the company's overall BI strategy. It should be the responsibility of the chief financial officer (or key users identified by the CFO) to represent the business and work with IT to ensure that performance management systems are deployed successfully.
Timo Elliott is senior director of strategic marketing at Business Objects, a San Jose-based developer of business intelligence solutions. He can be reached at firstname.lastname@example.org.
Sarah A. Blanchard, Business Objects Account Executive, Eastwick Communications, 1735 Technology Drive, Suite 430, San Jose, CA 95110, gave permission to use this article at DSSResources.COM on Wednesday, October 12, 2001. For more information check http://businessobjects.com/. Founded in 1990, Business Objects has over 13,100 customers in more than 80 countries. This article was posted at DSSResources.COM on October 28, 2001.